A spreadsheet with five years across the top. Revenue growing a comfortable percentage a year, food cost holding at the benchmark, labor holding at the benchmark, a net margin at the bottom that makes the deal work.
It exists because a bank asked for it, or an investor did, or a course told the operator to build one. And then something happens that nobody intends: the projection becomes the plan, and the plan starts running the building.
What A Pro Forma Actually Is #
A pro forma is a set of assumptions arranged to produce a required conclusion.
Read that again, because it is not cynicism. It is the mechanics. The document is built backward from an outcome — a loan that clears, an investor who commits, a deal that pencils. Every cell is then filled with a number that supports it.
Where do those numbers come from? Industry averages, a template’s defaults, and the comparable operations somebody found. Which means the model is built out of the same benchmarks that were never about your building, compounded across sixty months.
The inference: the model shows the operation works; therefore the operation works.
The Assumptions Nobody Tests #
Revenue ramp. Almost every template assumes a steady climb. Real operations do not ramp steadily. They have an opening period that is not representative in either direction, then a real level, then whatever your read and your market produce.
Cost percentages holding flat. The model runs food cost at a constant across five years. Your actual cost basis moves with supply, with your mix, with your volume, and with what your building is chosen for. Percentages that hold flat on a spreadsheet are an artifact of the spreadsheet.
Labor as a percentage. The heaviest and most dishonest cell in the document. Labor is dollars — specific people at specific rates for specific hours holding a specific standard. Expressed as a percentage it becomes an adjustable input, and when the model needs a better bottom line, the percentage gets lowered. Nobody has decided which hour comes out. That happens later, in the building, to a real shift.
Capacity treated as elastic. Growth in the model does not check whether the room, the kitchen, the hours, or the cast can produce it. Revenue growth on a spreadsheet requires no physical capability. In a building it requires all of it.
Nothing breaks. No equipment failure, no key person leaving, no bad period, no supply shock, no road construction. The absence of variance is what makes the bottom line clear enough to fund.
Owner compensation. Frequently understated or omitted, which makes the model work while making the operator’s actual life not work.
The Real Damage #
The document is built for an audience. Then the audience approves it, and the model stays.
Now it is the plan. Monthly performance gets read against the projection instead of against the building. A period comes in under the model and the correction is aimed at the variance rather than at what the operation actually needs. The operator is managing to a spreadsheet built out of strangers’ averages before he had ever run a shift in that room.
And the model outranks him. He sees something in his building that calls for a different decision, and the projection says no, because the projection is what the money agreed to. A document assembled to close a deal is now governing an operation it never described.
The Legitimate Use #
Two, and they are real.
Capital requirement. What has to be in the bank before the doors open and how long the operation can survive producing less than it needs. This is the most useful thing a model does and the most commonly underbuilt. Build it deliberately pessimistic.
Structural feasibility. Whether the deal can work at all, at rents and wages that exist, in a room of that size, at prices your market will actually pay. If it cannot work on honest numbers, the read stops there and that is the model earning its keep.
Everything past those two is a projection of a business nobody has run yet.
What Replaces It #
Work backward from what the building has to produce. Fixed dollars per period. The cast at what it actually takes. Maintenance and replacement. Reinvestment. What the operation owes you. Add it up. That is the number, and it came from your building instead of from a survey.
State every assumption as a sentence, with its source. Not a cell. A sentence. This many covers at this average check because of this specific read of this market. If a sentence cannot be written, the cell is a guess wearing a number.
Model in dollars, not percentages. Specific people, specific hours, specific product cost. Percentages hide what a change actually does to a shift.
Check every revenue assumption against physical capacity. Seats, hours, kitchen throughput at your peak, and how many hands it takes. Revenue the building cannot physically produce is not a projection.
Run the honest downside. What happens at seventy percent of your expectation, for longer than you would like. If the answer is that the operation dies, the capital plan is the thing to fix, not the projection.
Re-derive it every time the load moves. New rent, new daypart, a menu change, a volume shift, a cast change, a supply change. The numbers were drawn against a load, and when the load moves the model describes an operation you no longer have.
Read against your own prior periods, not against the model. Your operation last period is the only valid comparison. The model is what somebody needed to see before you opened.
What Changes Tomorrow #
Open whatever projection you are currently working against and go cell by cell on the assumption rows. Next to each one, write the sentence behind it and where the number came from.
Most rows will not produce a sentence, and the ones that do will mostly trace to an average, a template default, or a comparable operation somebody found. That inventory is the finding: the document running your operation was assembled out of numbers about other people’s buildings.
Then build the one number that is actually yours. Fixed dollars, cast at real cost, maintenance, reinvestment, what the operation owes you. Divide by the covers your book actually produces. That is required margin per Guest, derived from your building.
Set it next to what your last period actually produced per Guest. That gap is the only projection you need this week, and unlike the spreadsheet, it is about your operation.
Digging Deeper #
Every term in my framework lives in the Knowledge Base: kb.jeffreysummers.com
Terms used here: [Measurement Lock-In], [Cover Blindness], [Cost Basis Opacity], [Static Decline], [Sameness Machine], [Default Architecture], [Designed Architecture], [Differentiation Economics], [Structural Scale], [Outcomes Formula], [Customer Architecture], [Constraint Architecture], [Constant Expiry], [Transactional Cost-Plus]
More of my work on this, all of it in the Knowledge Base:
- 5.X Undercapitalization Is A Knowledge Problem — the real reason the model failed
- 5.X The Empty Chair — the cost the projection never carries
- 5.MQ.0 The Million Dollar Question — the number to work backward from
- 5.X Your P&L Is A Story — reading what actually happened
- 4.X The Opening Model Is Broken — the assumptions the template inherited
- 037 Spreadsheet Vs Dining Room — which one is the operation
